From: William Donzelli <>
Date: Mon Jan 12 21:26:28 1998

> As you pointed out in this and a subsequent post there were many "solid state
> electronic" calculators available by 1971. Hey, for that matter IBM's S/360
> shipped transistorized general purpose transistorized computers (running DOS
> and TOS among others) by 1964, and by 1971 they were even beginning to
> incorporate integrated circuitry into what would become S/370 computers.
> Interestingly the architecture (or its modern desecndant) was not put on a
> single microprocessor until just a few years ago (1995 saw the 3490 CMOS
> mainframe on a chip).

This is mostly because of the need for speed in an very complex
system. They probably could not have put a S/360 implemenation on one chip
and still get decent performance until recently. There is quite a bit of
extra circuitry in the big IBMs that nobody ever sees - mostly in the
redundancy and error checking realms missing in microprocessors.
For example, the memory interface for the older (and probably the newer,
as well) RS/6000s has SECDED error handling, memory scrubbing, and faulty
bit substitution all built in. AS/400s are the same way. In the mainframe
line, this philosophy goes even further, with things like parity on
internal busses. In other words, it is no suprise that a 3081 (ca. 1983)
is implemented using over a thousand gate arrays. The amazing thing is
that IBM was able to get all of those _hot_ chips working in such a small
space (about a cubic foot, maybe two).

Yes, I lust for a S/360 or 370, and would even settle for a 3033 or 3081.

William Donzelli
Received on Mon Jan 12 1998 - 21:26:28 GMT

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:30:56 BST